Simple and powerful site deployment with capistrano Kim Pepper [email protected] @scorchio96 d.o: kim.pepper How are you deploying now?
• FTP, SFTP, SCP? • SSH + GIT? • Aegir? • Drush? • Debian packages? • Don't know - operations team does it for me! What are the issues? • lots of manual processes in deployment o backup db o update code o run updatedb o flush caches • potential for human error
Small organisation? • too loose server access and permissions
Large organisation? • too restrictive permissions requiring sysadmin or super user to deploy
You didn't know you had a problem?
A great tool for web ops needs to be automatic, accessible and actionable” – @briandoll
Sysadmins don't want you touching their stuff Don't cross the line
• Sysadmins don't want you touching their stuff
Standard server environments: • conventions • consistency • security • monitoring • quality control • automated configuration management tools (e.g. puppet, chef) Working together to get the job done Working together to get the job done
Ideally:
• Developers are able to deploy rapidly • Sys admins have tight control of environments • All manual processes are automated Capistrano Capistrano
Capistrano is a developer tool for deploying web applications. It is typically installed on a workstation, and used to deploy code from your source code management (SCM) to one, or more servers. Server Directory Structure
• Each deployment lives in its own time-stamped directory o e.g. 20111101020759/ • settings.php and files directory are shared across deployments, and don't live in version control • a 'current' symlink points to the current release, and is used as the site root in apache config Server Directory Structure e.g. under /var/www/example.com/ current -> 20111101020759 (symlink to current release) releases/ 20110929042332/ 20110929045734/ 20111101001651/ 20111101020759/ shared/ cached-copy/ files/ settings.php
Symlinks for shared files
Symlinks are made back to the shared folder, so they are not lost between deployments, e.g.: sites/default/files -> shared/files sites/default/settings.php -> shared/settings.php
How does it work? • No software installed on the server, just locally • Capistrano 'builds' commands to be executed remotely • executed on the server over ssh
Executing a deploy It's a simple one-line command to deploy from your local machine: cap deploy
Deployment Workflow Simple Deployment Workflow Hooks • before and after hooks at each stage of the deployment lifecycle • extend or customise default capistrano behaviour
Examples: • deploy • deploy:setup • deploy:symlink • deploy:update_code • deploy:rollback
Rolling-back a deploy Made a mistake? Rollback with: cap deploy:rollback
• Sets the 'current' symlink to the previous release directory • No automatic re-import from backed up database
Full list of available tasks with:
cap -T Set-up and Configuration What is it written in?
Ruby! (I know, it's not made with Drupal).
• Originally designed for deploying Rails apps • Cross-platform • Many OS's have it already (e.g. Mac OSX) • RubyInstaller for Windows users http://rubyinstaller.org/ • Rubygems installed from http://rubygems.org/ RubyGems you'll need
• capistrano • capistrano-ext • railsless-deploy • capistrano-drupal gem install capistrano capistrano-ext \ railsless-deploy capistrano-drupal
A Drupal Helper Gem + = capistrano-drupal gem http://rubygems.org/gems/capistrano-drupal
• performs common Drupal deployment tasks • also calls Drush commands on the server • implements before and after hooks deploy:setup • Creates files directory and sets correct group- writeable permissions drupal:symlink_shared • re-creates symlinks to settings.php and files directory • called after "deploy:symlink" Capistrano-drupal gem cont... drush:backupdb • Calls the drush backup-migrate command to create a db backup • called before "drush:updatedb" drush:updatedb • Calls the drush updatedb command • called after "deploy:symlink" drush:cache_clear • Calls the drush cc all command to clear the Drupal caches • called after "deploy:symlink" Drupal deployment workflow Using Capistrano in Your Project Site Directory Structure Drupal site root lives under an app directory so it stays separated from the deploy config directory. e.g. locally under ~/Sites/example.com/ app/
# example drupal site set :application, "example.com" set :scm, "git" set :repository, "[email protected]:example/example.git" set :branch, "master" set :port, 22 set :deploy_to, "/var/www/#{application}" set :user, "exampleuser" set :runner, "exampleuser" role :app, 'www.example.com' Optional Extras Server Roles • Support for multiple servers o e.g 2 app servers, 1 db server • Specified as roles: role :app, 'www.example.com' role :db, '10.1.1.10'
• Capistrano tasks can be run only on servers with specific roles (e.g. only deploy code on app server)
Password-less Deploy • Use ssh public keys • Less manual process involved • Uses ssh key-forwarding for accessing github repo
Multi-Stage Deployments • Have different deployment configuration for dev, staging, and production environments (or more) • e.g. different hosts, git branches, root directory • config/ deploy.rb deploy/ dev.rb staging.rb prod.rb
Deploy using a stage name: cap prod deploy
Custom commands Add your own 'tasks' by defining them in deploy.rb.
This will echo 'hello' on the server after updating code.
Filtering Add filters to specific 'roles' or hosts: cap deploy HOSTS=app2.server.hostname cap deploy ROLE=app Gateways You can set up a gateway server to tunnel deployments through: set :gateway, '[email protected]:2222' Git Submodules You can enable capistrano to automatically fetch updates for git-submodules on deploy Demo Questions? Kim Pepper [email protected] @scorchio96 d.o: kim.pepper